A first stage, the paper can be read in English, fully understand the algorithm described paper inside. We know the merits and applicable scenes algorithms, and can use their own familiar language out of the algorithm. The project regarded inside the class level, and should be able to get high wages.
The second stage, and the paper can not be fully appreciated that algorithms, but known manner effects and tuning algorithms in certain scenarios. Can be found right on GitHub open source implementation of the above, carried out using even optimized. If the project has a wealth of experience and practical operation experience will be a plus.
Third grade, totally do not understand English paper, know the algorithms used, but can not distinguish the pros and cons of various algorithms and application scenarios, the algorithm can be assembled using python, C ++ and other languages libraries.
Everyone reading the above description, feel, data, what is the core competitiveness of AI? The syntax is python, python or library? Of course not, I think the core or the algorithm itself, but to varying degrees of understanding of the algorithm, will be at different levels. Here reading ability in English for many people, will become a hurdle.
Currently self-study and training of students to participate in most of the third stage, the third stage did not even reach. There are training students to come contact me. I asked him at present is how to learn. He said watching training videos, books too difficult, not read. If this is the case, IMHO, might never have a chance advanced to the second grade. Because the need to look behind the books, only harder than the entry books.